Comparing two different family-based childhood obesity treatment programmes in a medically underserved region: Effectiveness, engagement and implementation outcomes from a randomized controlled trial.

Background: Access to evidence- and family-based childhood obesity (FBCO) treatment interventions is a challenge, especially in underserved regions where childhood obesity disparities persist.

Objective: Compare two 6-month FBCO treatment interventions, iChoose (high intensity, parent-child dyads) and Family Connections (low intensity, parents only), in one underserved US region.

Methods: This unblinded, RCT reports on effectiveness and implementation outcomes. Eligibility included children ages 5-12 with BMI ≥85th percentile. Analyses included descriptive statistics and intention-to-treat Heckman treatment effect models.

Results: Enrolled children (n = 139, mean age 10.1 ± 1.7 years, 30% with overweight, 70% with obesity, 45% black, 63% Medicaid) were randomly assigned to iChoose (n = 70) or Family Connections (n = 69). Retention rates were 63% for iChoose and 84% for Family Connection. Among children, 6-month BMI z-score changes were not statistically significant within iChoose [BMI z-score 0.03 (95% CI = -0.13, 0.19)] or Family Connections [BMI z-score 0.00 (95% CI = -0.16, 0.16)]. Likewise, parents' BMI changes were not statistically significant. No adverse events were reported. Both programmes were delivered with high fidelity (77%-100%). Engagement in core components was 25%-36% for iChoose and 52%-61% for Family Connections. Implementation costs per child with improved BMI z-score were $2841 for iChoose and $955 for Family Connections.

Conclusions: Neither intervention yielded significant improvements in child BMI z-score or parent BMI, yet both were delivered with high fidelity. Relative to iChoose, descriptive data indicated higher retention, better engagement, and lower costs for Family Connections-suggesting that a lower intensity and parent-focused programme may better fit the intended audience's context.